The Digital Freight Train is expected to revolutionise the rail freight industry. A critical aspect of this transformation is real-time condition monitoring of air brake systems, which are among the leading causes of train malfunctions. To achieve this goal, advanced algorithms for air brake modelling are required. This paper introduces a computationally efficient air brake model tailored for real-time diagnostic applications. A hybrid approach, integrating both empirical data and simplified fluid-dynamic equations, has been adopted. Compared to other air brake models found in the literature, the innovative contributions of the presented model are the reduction of the number of required parameters and the estimation of the brake cylinder pressure directly from the main brake pipe pressure using a feed-forward approach. Moreover, a new approach in the evaluation of the first braking phase and the brake cylinder pressure build-up as the saturation of the brake mode is presented. The model input includes the main brake pipe pressure, the weighing valve pressure, and the brake mode, and the output includes the pressure at the brake cylinder. The air brake model has been validated using data from a previous experimental campaign. The model’s accuracy in replicating the air brake system mechanism makes it well-suited for future development of model-based algorithms designed for air brake fault detection.
Read full abstract